Isolation, identification, and synthesis of two oxidative degradation products of olanzapine (LY170053) in solid oral formulations

Article abstract of DOI:10.1002/jps.21146

Two impurities found in both stressed and aged solid-state formulations of olanzapine have been identified as (Z)-1,3-dihydro-4-(4-methyl-1-piperazinyl)-2- (2-oxopropylidene)-2H-1,5-benzodiazepin-2-one (1) and (Z)-1-[1,2-dihydro-4-(4- methyl-1-piperazinyl)-2-thioxo-3H-1,5-benzodiazepin-3-ylidene]propan-2-one (2). The structures indicate that the two impurities are degradation products resulting from oxidation of the thiophene ring of olanzapine. The impurities were isolated by preparative HPLC from a thermally stressed formulation, and characterized by UV, IR, MS, and NMR. A synthetic preparation of compounds 1 and 2 by reaction of olanzapine with the singlet oxygen mimic 4-phenyl-1,2,4- triazoline-3,5-dione (PTAD) is presented. The structure of 2 was also determined by single-crystal X-ray diffraction analysis. A degradation pathway for the formation of 1 and 2 is proposed.

Full text of DOI:10.1002/jps.21146

Isolation, Identification, and Synthesis of Two Oxidative
Degradation Products of Olanzapine (LY170053) in
Solid Oral Formulations
STEVEN W. BAERTSCHI, HEIKO BRUNNER, CHARLES A. BUNNELL, GARY G. COOKE, BENJAMIN DISEROAD,
DOUGLAS E. DORMAN, PATRICK J. JANSEN, CRAIG A.J. KEMP, STEVEN R. MAPLE, KAREN A. McCUNE,
JEFFREY L. SPEAKMAN
Eli Lilly and Company, Analytical Sciences Research and Development, Lilly Corporate Center, Indianapolis, Indiana 46285
Received 10 December 2006; revised 18 March 2007; accepted 20 March 2007
Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21146
ABSTRACT: Two impurities found in both stressed and aged solid-state formulations of
olanzapine have been identified as (Z)-1,3-dihydro-4-(4-methyl-1-piperazinyl)-2-(2-oxo-
propylidene)-2H-1,5-benzodiazepin-2-one (1) and (Z)-1-[1,2-dihydro-4-(4-methyl-1-
piperazinyl)-2-thioxo-3H-1,5-benzodiazepin-3-ylidene]propan-2-one (2). The structures
indicate that the two impurities are degradation products resulting from oxidation of the
thiophene ring of olanzapine. The impurities were isolated by preparative HPLC from a
thermally stressed formulation, and characterized by UV, IR, MS, and NMR. A synthetic
preparation of compounds 1 and 2 by reaction of olanzapine with the singlet oxygen
mimic 4-phenyl-1,2,4-triazoline-3,5-dione (PTAD) is presented. The structure of 2 was
also determined by single-crystal X-ray diffraction analysis. A degradation pathway for
the formation of 1 and 2 is proposed. ß 2007 Wiley-Liss, Inc. and the American Pharmacists
Association J Pharm Sci 97:883–892, 2008
Keywords: olanzapine; LY170053; drug–excipient interactions; degradation; degra-
dation pathways; isolation; drug oxidation; impurities; structure elucidation; HPLC;
stress testing; thiophene; single crystal X-ray diffraction
INTRODUCTION
degradation products were observed. Two of these
degradation products, compounds 4 and 5, were
drug substance degradation products that had
been previously identified during drug substance
stress testing studies; compound 5 is also a human
metabolite. The other two degradation products
(1 and 2), detected in both an aged tablet formu-
Olanzapine, a thienobenzodiazepine with a struc-
ture as shown in 3, is a marketed antipsychotic
agent for treatment of both the positive and
negative symptoms of schizophrenia. During the
development of this compound, stress testing
and stability studies were conducted on tablets
and related formulations to determine potential
degradation products, to develop stability indu-
cing analytical methods, and to help ensure the
safety and potency of the drug product. In stress
testing studies of different formulations, four
lation and
a granulated formulation, were
observed to increase over time upon exposure
to thermal stress (e.g., 40–608C). These unknowns
had not, however, been observed to form
during solid-state stressing of the drug substance
at elevated temperatures (e.g., 70 or 858C)
(Baertschi et al., unpublished results) in the
absence of excipients. Because of the potential
(during storage and distribution) for the level
of these two degradation products to approach
the International Conference on Harmonization
Correspondence to: Patrick J. Jansen (Telephone: 317-276-
1423; Fax: 317-276-1388; E-mail: patjansen@Lilly.com)
Journal of Pharmaceutical Sciences, Vol. 97, 883–892 (2008)
ß 2007 Wiley-Liss, Inc. and the American Pharmacists Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
883

884
BAERTSCHI ET AL.
(ICH) impurity threshold levels,1 identification
studies were undertaken. This report describes
the isolation, structure elucidation, and synthesis
of these two degradation products.
Fast-Atom Bombardment (FAB)-MS
FAB-MS and FAB-MS/MS data were obtained
using a VG ZAB-3 (B1EB2) triple-sector mass
spectrometer or a VG ZAB-2SE (BE) two-sector
mass spectrometer. Samples were dissolved in
‘‘Magic Bullet’’ (a 5:1 solution of dithiothreitol/
dithioerythritol in methanol) and bombarded with
8 keV Xenon atoms in the ZAB-3 or Cs ions having
a net energy of 12 keV in the ZAB-2SE.
EXPERIMENTAL
NMR
Proton and carbon-13 NMR spectra were recorded
at 299.957 and 75.432 MHz, respectively, using a
Varian Unity spectrometer equipped with a 5-mm
Nalorac indirect detection probe and a MatrixTM
shim system. Samples consisted of 11.7 mg/mL of
1 and 9.4 mg/mL of 2 dissolved in dimethylsulf-
oxide-d6. Chemical shifts are reported in parts per
million relative to dimethylsulfoxide-d6 (2.50 ppm
1H, 39.50 13C). All spectra were recorded in a
nonspinning mode.
HPLC/MS
Pneumatically assisted electrospray ionization
(ESI) mass spectrometric analyses were per-
formed using a triple-quadrupole tandem mass
spectrometer (Micromass Quattro I, Beverly MA).
Operating in the positive ion mode, the spray
needle was held at a potential of þ2.75 kV. The
skimmer cone potential, which affects the degree
of collisional activation in the source, was set to
À25 V. The source temperature was maintained at
808C. The 1 mL/min flow rate from the HPLC
column was split by means of a low dead volume
tee. The waste line was connected to a variable
back-pressure flow controller which was adjusted
to allow a flow of approximately 50 mL/min
through the capillary connected between the tee
and the mass spectrometer. Spectra were collected
Electron Impact (EI)-MS
Positive ion 70 eV EI-MS data were obtained using
a VG/Fisons 7070E (Beverly, MA) double focusing,
forward geometry magnetic sector mass spectro-
meter by means of a direct-insertion heated probe
for sample introduction.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
DOI 10.1002/jps

ISOLATION, IDENTIFICATION, AND SYNTHESIS OF TWO OXIDATIVE DEGRADATION PRODUCTS
885
at a rate of one every 0.7 s while scanning from m/z
150 to m/z 550.
Mobile phase B was increased linearly from 0 to
100% at 3.33% per minute for 30 min.
Preparative HPLC
IR
Preparative HPLC separations were carried out
using a Rainin HPXL solvent delivery system
equipped with a Dynamax UV detector (Model
UV-1) and a Dynamax fraction collector (Model
FC-1). The column, a 50.8 Â 250 mm Kromasil
KR100-10C18, 10-mm particle size (Technikrom),
was operated at ambient temperature. The flow
rate was maintained at 70 mL/min and detection
was at 220 nm. Mobile phases A and B consisted of
0.1% trifluoroacetic acid in water and 0.1%
trifluoroacetic acid in ACN, respectively. The
solvent composition was held at 13% mobile phase
B for 10 min. Mobile phase B was then increased
linearly from 13 to 30% at 0.85% per minute for
20 min. The fractions corresponding to 1 and 2
were collected into dry ice cooled flasks. The
solvent was removed from the fractions by
lyophilization. Transfer of the resulting com-
pounds to vials was facilitated by dissolving in
methanol. The methanol was removed by eva-
poration with a stream of nitrogen and gentle
heating (ꢀ308C).
Samples were prepared as micro-KBr pellets.
Spectra were obtained on a Nicolet FTIR, model
5SXC, with a DTGS KBr detector. Data were
collected between 400 and 4000 cmÀ1, with a
resolution of 4.0 cmÀ1. A total of 1024 scans were
obtained, and processed using Happ-Genzel apo-
dization with a gain of 8.0.
(1): 3199, 3417 cmÀ1 (N–H, broad); 3036 cmÀ1
(aromatic C–H); 2922, 2975 cmÀ1 (C–H stÀre1tch);
1678 cmÀ1 (C O stretch, strong); 1638 cm (C
–
–
–
–
C, weak); 1581, 1609 cmÀ1 (aromatic ring,
medium). (2): 3180, 3429 cmÀ1 (N–H, broad);
3006, 3088 cmÀ1 (aromatic C–H); 2935, 2963 cmÀ1
(C–H stretch); 1676 cmÀ1 (C O stretch, strong);
–
–
1635 cmÀ1 (C C, weak); 1578, 1608 cmÀ1 (aromatic
–
–
ring, medium); 1520 cmÀ1 (C S, weak).
–
–
Analytical HPLC
HPLC analyses were carried out on a Waters
photodiode array (PDA) system consisting of a
600E controller and pump, a 715 Ultra Wisp, and
a 996 PDA detector. The UV spectra were obtained
on the compounds as they eluted from the
chromatographic column.
Solid-Phase Extraction
Solid phase extractions were carried out using
Varian 10 g, Mega Bond-Elut C18 cartridges.
Stressed olanzapine granule samples were dis-
solved in 0.1% TFA in water, applied to the SPE
cartridge, washed with water, and eluted with
methanol.
Ion-Pairing Method
The column, a 4.6 Â 250 mm Zorbax Rx-C8, 5-mm
particle size (MACMOD Analytical), was operated
at ambient temperature. The flow rate was
maintained at 1.0 mL/min. Mobile phases A and
B consisted of 52:48 and 30:70 (v/v) 30 mM SDS, 50 mM
H3PO4 (pH adjusted to 2.5 with NaOH)/acetoni-
trile, respectively. The solvent composition was
held at 100% mobile phase A for 3 min. Mobile
phase B was then increased linearly from 0 to
100% at 5.9% per minute for 17 min and held at
100% B for 20 min.
X-ray Crystallographic Analysis
X-ray diffraction analysis was carried out on a
Siemens R3m/V diffractometer at room tempera-
ture. A crystal was monoclinic, space group C2/c,
˚
˚
3
with a unit cell a ¼ 28.403 (5)A, b ¼ 9.775 (2)A,
˚
˚
c ¼ 17.923(4)A, b ¼ 128.45(2)8, V ¼ 3897.2 (13)A ,
Z ¼ 8, Dx ¼ 1.297 g/cm3, m ¼ 1.716 mmÀ1. CuKa
LC/MS Compatible Method
˚
radiation (l ¼ 1.54178 A). A total of 2659 reflec-
The column used for the ion-pairing method was
also used for this method and was operated at
ambient temperature. The flow rate was main-
tained at 1.0 mL/min. Mobile phases A and B
consisted of 95:5:0.1 and 20:80:0.1 (v/v/v) water/
acetonitrile/trifluoroacetic acid, respectively.
tions with F > 4.0s (F) gave R ¼ 0.0774 with 242
parameters refined. The structure was elucidated
by direct methods (Siemens SHELXTL PLUS
(VMS)). Final refinement of the nonhydrogen
atoms was done by full-matrix least-square
refinement.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008

886
BAERTSCHI ET AL.
Synthesis of 1 and 2
(99.7% by area HPLC) suitable for single-crystal
X-ray analysis.
Preparation of 4-Phenyl-1,2,4-triazoline-3,5-dione
(PTAD) Adduct
Formulations
Olanzapine (3.0 g) was dissolved in ethyl acetate
(750 mL) at 20–258C and PTAD (2.2 g, Aldrich,
Milwaukee, WI), was added. A tan precipitate
formed, and the mixture was stirred for 10 min.
Tablets were obtained from Pharmaceutical
Product Development, Lilly Research Labora-
tories (Indianapolis, IN). The tablet formulation
was composed of the following ingredients:
olanzapine, lactose, hydroxypropyl cellulose, cros-
povidone, microcrystalline cellulose, and magne-
sium stearate. A granulated formulation was
obtained from Lilly-Hamburg (Hamburg, Ger-
many). The granulated formulation was composed
of the following ingredients: olanzapine, hydro-
xypropylmethyl cellulose, mannitol, and polysor-
bate 20.
Preparation of 1
To the above precipitate in ethyl acetate was
added 0.1% aqueous trifluoroacetic acid (750 mL).
The biphasic mixture was stirred for 5 h at
ambient temperature and the pH was adjusted
to about 6.5 with 6% aqueous NaHCO3 solution
(ꢀ25 mL). The mixture was filtered through hyflo
to remove insoluble material and the layers were
separated. The aqueous layer was extracted with
ethyl acetate (200 mL) and then twice with
dichloromethane (250 mL). The combined organic
layers were dried over magnesium sulfate and the
solvent removed by rotary evaporation to give a
crude residue of 1, 0.27 g (75% by area HPLC
analysis, 7% yield from olanzapine).
RESULTS AND DISCUSSION
Generation, Isolation, and Spectroscopic
Characterization of 1 and 2
Olanzapine tablets and a granulated formulation
(see Experimental for a listing of excipients) were
exposed to a variety of stress conditions. Analysis
by HPLC with PDA-UV detection revealed that at
elevated temperature and low humidity two new
peaks (1 and 2) were forming at low levels in both
the tablets and granulated formulation (see
Fig. 1). For example, after 6 months storage of
the tablets at 408C in a closed container, the levels
of 1 and 2 were in the range of 0.1–0.2%, while
levels in the granulated formulation were typi-
cally 1–2% after 3 months at 608C. To obtain
molecular weight information, an LC-MS-compa-
tible method was developed (see Fig. 1 for HPLC
chromatogram and Fig. 2 for UV spectra of
olanzapine, 1, and 2). LC-electrospray/MS ana-
lyses of these samples (data not shown) estab-
lished the nominal molecular weights of these
compounds (312 and 328 amu for 1 and 2,
respectively). These molecular weight results
suggested that 2 might be a simple oxidation
product of olanzapine (M þ 16 amu), while 1
appeared to be an isomer of olanzapine.
Preparation of 2
To the mixture of olanzapine/PTAD adduct in
ethyl acetate was added 0.1% aqueous trifluor-
oacetic acid (750 mL). The biphasic mixture was
stirred for 5–10 min and the pH was adjusted
to about 6.5 with 6% aqueous NaHCO3 solution
(ꢀ20 mL). The mixture was filtered through hyflo
to remove insolubles and the layers were sepa-
rated. The ethyl acetate layer was dried with
magnesium sulfate and the solvent removed by
rotary evaporation to give a crude residue of 2,
1.98 g, (33% by area HPLC analysis, 22% yield
from olanzapine).
Purification of 2
The crude residue of 2 (12.5 g) was dissolved in
methanol (250 mL) and purified by column
chromatography (1 kg of silica gel in a 4.5 in
diameter column) eluted with 2:1 v/v methanol/
ethyl acetate). The appropriate fractions were
collected from two independent runs and concen-
trated to ꢀ100 mL by rotary evaporation.
Compound 2 crystallized from this solution at
À108C as the methanolate dihydrate. The yellow-
gold crystals were collected by filtration and dried
in vacuo to give 2.07 g (19% yield recovery, 4%
yield from olanzapine) of 2 in a form and purity
To facilitate further spectroscopic characteriza-
tion, larger amounts of compounds 1 and 2 (along
with olanzapine) were extracted from a sample
of the granulated formulation (which had
been stressed at 708C for ca. 2 months) using
solid-phase extraction. The compounds were
isolated as the trifluoroacetate salts using
preparative reversed-phase HPLC (see Experi-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
DOI 10.1002/jps

ISOLATION, IDENTIFICATION, AND SYNTHESIS OF TWO OXIDATIVE DEGRADATION PRODUCTS
887
tively. The molecular formula for 1 showed that it
was not an isomer of olanzapine, since there was
no sulfur atom but rather two additional oxygen
atoms (as compared to olanzapine). Compound 2
corresponded to the molecular formula of olanza-
pine plus one oxygen atom.
The proton and carbon-13 NMR spectra of the
isolated compounds are consistent with 1 and 2
being structurally similar. For ease of discussion
the NMR data for 1 will be described and the
corresponding assignments can be made for 2.
See Table 1 for tabulated NMR data. The NMR
data indicate the presence of a 1,2-disubstituted
benzene structure in both compounds. Two- and
three-bond heteronuclear correlations identify the
resonances of carbons 5A and 9A and confirm this
substructure. The resonance at 6.29 ppm in the
proton spectrum of 1 can be assigned to position 3
based upon the long-range coupling observed
between this proton and the carbons at positions
10A, 3A, 4, and 2. The chemical shifts of the carbons
at positions 10A and 2 (165.25 and 196.82 ppm)
are consistent with the attachment of oxygen
atoms at these positions. The nonprotonated
carbon at position 3A also shows long-range
coupling to the NH proton at position 10, confirm-
ing its assignment. The methyl group attached at
position 2 is assigned based upon the two-bond
correlation observed between the methyl protons
and the carbon at position 2. This leaves only the
assignment of the N-methyl piperazine substruc-
ture to complete the molecule; however, while the
presence of the N-methyl piperazine ring is
indicated by the mass spectral data (see below)
not all of the resonances were observed for the N-
methyl piperazine substructure in the NMR
spectra. The absence of some of the N-methyl
piperazine resonances has been observed for other
olanzapine-related compounds and has been
ascribed to a broadening of the resonances due
to a combination of rotation around the central
Figure 1. HPLC chromatograms (UV, 220 nm) show-
ing degradation profiles of an olanzapine tablet stressed
at 708C for 53 days. Upper chromatogram: ion-pairing
HPLC method. Lower chromatogram: LC/MS-compati-
ble HPLC method.
mental). Accurate mass FAB-MS measurements
for
1 and 2 indicated molecular formulas
of C17H20N4O2 (observed [M þ H]þ 313.1670,
calculated 313.1665) and C17H20N4OS (observed
[M þ H] 329.1434, calculated 329.1436), respec-
–
amidino C–N bond of >N–C N, piperazine chair–
–
chair interconversion, and nitrogen inversion.2 In
order to sharpen the broadened NMR resonances,
the free base of 1 was prepared by dissolving in
aqueous base and extracting with chloroform. The
NMR spectra of the extracted material indeed
showed a sharpening of the methylene reso-
nances, and the chemical shifts of 20/60, 30/50,
and 40 N–CH3 were very similar to those observed
for olanzapine.3 An attempt to prepare the free
base of 2 in the same manner was unsuccessful
due to the rapid degradation of 2 to 1 under basic
conditions. The only significant difference in the
Figure 2. UV spectra of olanzapine and impurities 1
and 2. Spectra were obtained as the compounds eluted
from the HPLC column using UV-PDA (ion-pairing
HPLC method).
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008

888
BAERTSCHI ET AL.
Table 1. NMR Chemical Shift Assignments for 1 and 2
1
2
Site
Atom Type
dH (ppm)
dC (ppm)
LRCa
dH (ppm)
dC (ppm)
LRCa
–
2
>C O
—
196.82
30.18
128.08
137.45
151.97
—
129.29
123.32
124.53
126.92
121.97
138.79
—
2-CH3, 3
—
196.97
30.41
124.31
141.34
152.09b
—
131.06
122.09
127.33
126.27
122.89
140.74
—
2-CH3, 3
–
2-CH3
3
3A
4
5
5A
6
7
8
9
9A
10
10A
20, 60
30, 50
N–CH3
–CH3
CH–
C<
C<
NH–
C<
CH–
CH–
CH–
CH–
C<
2.20
6.29
—
—
10.17
—
7.01b
7.05b
7.05b
7.10b
—
2.20
6.09
—
—
10.11
—
7.26b
7.10b
7.16b
7.04b
—
–
–
–
–
3, 10
3
10
–
–
–
–
–
7
7
8
9
6
7
6
–
–
–
7 or 8
9
6
10
–
–
–
–
–
–
–
–
6, 9, 10
–NH–
10.75
12.79
a
–
>C X
—
165.25
3
—
190.71
3
–
c
c
c
c
–CH2–
–CH2–
–CH3
3.54b
2.87
51.96b
42.19
N–CH3
N–CH3
c
c
2.86
42.22
a‘‘Long-Range Correlations’’, that is, carbon–proton correlations through two- and three-bond coupling. In each cell is a list of the
protons which are within two or three bonds of the carbon in that row.
bChemical shift taken from one-bond heteronuclear correlation experiment.
cResonance not observed.
carbon NMR spectra of 1 and 2 is in the chemical
shift of the carbon at position 10A, and the
difference supports the substitution of an oxygen
at this position in 1 and a sulfur in this position in
2. The configuration of the seven-membered ring
exocyclic double was determined from the X-ray
crystal structure of 2 (see below).
oxidation of the thiophene ring by suitable
oxidation reagents (such as peroxy acids) simul-
taneously oxidized the distal tertiary nitrogen of
the N-methyl piperazine ring to the N-oxide, even
when conducted at low pH. Oxidation using
radical initiators5 provided <10% yields of 1
and 2 and were complicated by the presence of
other products. For example, the best results
using radical initiators were obtained when
olanzapine was exposed to the radical initiator
VAZO 52 (2,2’-azobis(2,4-dimethylvaleronitrile))
in an air-purged solution of acetonitrile after 6 min
at 708C, yielding five main products: 2 (5.3%), 1
(3.1%), unknown 1 (4.5%), unknown 2 (4.0%),
unknown 3 (0.7%). Oxidation involving singlet
oxygen (using rose bengal as a photosensitizer)
provided reasonable yields of 2 along with several
other products, complicating the isolation and
purification procedures. Use of the singlet oxygen
mimic PTAD (6) provided the best results, and a
relatively simple procedure (see Experimental)
was developed which provided reasonable yields
after purification.
The IR spectra of compounds 1 and 2 also
support the structural similarity of these two
compounds (see Experimental section). The only
difference in the spectra of the two compoundsÀis1
–
the presence of a C S stretch at about 1520 cm
–
in the spectrum of 2 that is absent in the spectrum
of 1.4
EI spectra of 1 and 2, with the major fragments
depicted, are shown in Figure 3. Assignment of the
fragments (based upon accurate mass data)
clearly supports the presence of the N-methylpi-
perazine moiety in both structures.
Synthesis of 1 and 2 from Olanzapine
For the preparation of 1 and 2 in multi-gram
quantities at purities suitable for use as reference
material, a method of oxidative transformation of
olanzapine was sought that would provide reason-
ably high yields as the reaction was scaled up.
This transformation proved to be difficult, since
The transformation of substituted thiophene
rings into the desired keto/thioketo structure
using PTAD has been previously described by
Del Mazza.6 The mechanism of the reaction of
PTAD with alkenes has been investigated;
depending on substrate and conditions, the initial
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
DOI 10.1002/jps

ISOLATION, IDENTIFICATION, AND SYNTHESIS OF TWO OXIDATIVE DEGRADATION PRODUCTS
889
Figure 3. EI mass spectrum and assignment of key fragment ions of 1 (top) and 2
(bottom).
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008

890
BAERTSCHI ET AL.
Scheme 1. Proposed pathway for preparation of 1 and 2 from reaction of PTAD with
olanzapine.
step (see Scheme 1) can apparently involve either
a [4 þ 2] cycloaddition, a [2 þ 2] cycloaddition, an
intermediate aziridinium imide (AI)7, or direct
electrophilic attack of PTAD on the thiophene
ring. To determine which of these possibilities was
occurring, attempts were made to characterize the
precipitate that forms immediately upon reaction
of olanzapine and PTAD. Unfortunately, the
precipitate was a mixture of at least three
components by TLC and NMR. One of the
components was unstable in DMSO-d6 and
decomposed at ambient conditions over several
hours; the other components have not been
identified. Regardless of the initial step, rearran-
gement to a hydrazide (HYD), as suggested by Del
Mazza, followed by hydrolysis leads to
(Scheme 1).
2
X-ray Crystal Structure of 2
The X-ray structure of 2 offered confirmation of
the ketothiolactam structure instead of the
thioketone lactam isomer (Fig. 4). In addition,
the Z-configuration of the exocyclic double bond
was confirmed. The torsional angle for thiolactam
–
–
C S and exocyclic C C double bonds n(S10A-
–
–
C10A-C3A-C3) was shown to be almost perpendi-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
DOI 10.1002/jps

ISOLATION, IDENTIFICATION, AND SYNTHESIS OF TWO OXIDATIVE DEGRADATION PRODUCTS
891
Proposed Mechanism of Formation in
the Formulation
A proposed degradation pathway leading to 1 and
2 is represented in Scheme 2. The proposed
pathway involves radical-mediated oxidation
(autoxidation), in which either the thiophene ring
of olanzapine is attacked by a peroxy radical or a
hydrogen atom is abstracted from olanzapine to
form a delocalized radical that subsequently adds
molecular oxygen to the thiophene ring to form
hydroperoxide (a). As shown in the scheme, two
pathways from hydroperoxide (a) to hydroxyl (b)
can be envisioned, which leads directly to 2. The
formation of 1 from 2 involves a simple hydrolysis.
The proposed degradation pathway offers a
possible explanation for the empirical observa-
tions that compounds 1 and 2 do not form at
appreciable levels during solid state stress studies
of the drug substance. Excipients such as crospo-
vidone (tablet formulation) and polysorbate 20
(granule formulation) are known to undergo
Figure 4. Thermal ellipsoid representation of com-
pound 2 from single-crystal X-ray diffraction analysis.
cular (À82.58). Likewise, the C3A–C3 double bond
–
is highly bent from planarity with amidine C N
–
bond in the diazepine ring (nN5-C4-C3A-
C3 ¼ À101.08).
Scheme 2. Proposed degradation pathway leading to 1 and 2.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008

892
BAERTSCHI ET AL.
autoxidation processes to form peroxy radicals
and hydroperoxides over time upon exposure to
air.8–10 Thus, it is proposed that autoxidation
processes involving these formulation excipients
lead to low levels of peroxy radicals and peroxides
in the formulation over time, catalyzing the
formation of these degradation-related impuri-
ties. Formation of compounds 1 and 2 using the
radical initiator AIBN provides additional con-
firmation of the proposed radical-mediated degra-
dation pathway.
2. Chakrabarti JK, Hotten TM, Morgan SE, Pullar IA,
Rackham DM, Risius FC, Wedley S, Chaney MO,
Jones ND. 1982. Effects of conformationally
restricted 4-piperazinyl-10H-thienobenzodiazepine
neuroleptics on central dopaminergic and choliner-
gic systems. J Med Chem 25:1133–1140.
3. The observed proton/carbon-13 chemical shifts,
recorded in DMSO-d6, for 2 were 20/60: 3.57/44.57,
30/50: 2.32/54.06, 40 N-CH3: 2.18/45.24 ppm. The
corresponding olanzapine chemical shifts are: 20/
60: 3.33/46.43, 30/50: 2.38/54.48, 40 N-CH3: 2.21/
45.73 ppm.
4. Nakanishi K, Solomon P. 1977. Infrared Absorption
Spectroscopy. 2nd ed. Oakland, CA: Holden-Day
Inc., p. 50.
CONCLUSIONS
5. (a) Boccardi G. 2005. Oxidative susceptibility test-
ing. In. Baertschi SW, editor. Pharmaceutical
Stress Testing: Predicting Drug Degradation.
New York: Taylor & Francis, pp. 205–230. (b) Boc-
cardi G. 1994. Autoxidation of drugs: Prediction of
degradation impurities from results of reaction with
radical chain initiators. Il Farmaco 49:431–435.
6. Del Mazza DJ. 1982. Reaction of 25-dimethoxy-
thiophene with 4-phenyl-123-triazoline-35-dione.
J Chem Soc Chem Commun 8:1033–1034.
7. Smonou I, Khan S, Foote CS, Elemes Y, Mavridis
IM, Pantidou A, Orfanopoulos M. 1995. React-
ions of phenyltriazolinedione with alkenes. Stereo-
chemistry of methanol adducts to aziridinium imide
intermediates. J Am Chem Soc 117:7081–7087.
8. Chafetz L, Hong W-H, Tsilifonis DC, Taylor AK,
Philip J. 1984. Decrease in the rate of capsule
dissolution due to formaldehyde from polysorbate
80 autoxidation. J Pharm Sci 73:1186–1187.
9. Scheirs J, Bigger SW, Then ETH, Billingham NC.
1993. The application of simultaneous chemilumi-
nescence and thermal analysis for studying the
glass transition and oxidative stability of poly(N-
vinyl-2-pyrrolidone). J Polym Sci, Part B: Polym
Phys 31:287–297.
Two impurities found in aged or stressed formula-
tions of olanzapine have been identified as com-
pounds 1 and 2. The structures indicate that the
two impurities are degradation products resulting
from oxidation and ring-opening of the thiophene
ring of olanzapine. The formation of degradation
products 1 and 2 in stressed and aged solid-state
formulations is proposed to occur from autoxidation
processes catalyzed by formulation excipients.
Synthetic reaction of olanzapine with PTAD, a
singlet oxygen mimic, allowed preparation of the
two degradation products in reasonable yields.
ACKNOWLEDGMENTS
The authors thank Jennifer Olkowski for work
involving the crystal structure determination of 2,
and Ross A. Johnson for NMR analysis of the
Olanzapine-PTAD precipitate. The expert assis-
tance of Gregory A. Stephenson in processing of
the X-ray data is gratefully acknowledged.
10. Hartauer K, Arbuthnot G, Baertschi SW, Johnson
RA, Luke WD, Pearson N, Rickard E, Tsang P,
Wiens RE. 2000. Influence of peroxide impurities
in povidone and crospovidone on the stability of
raloxifene hydrochloride in tablets: Identification
and control of an oxidative degradation product.
Pharm Dev Technol 5:303–319.
REFERENCES
1. Guidance for Industry: Q3B(R2) Impurities in New
Drug Products, International Conference on Har-
monization, 2006.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 2, FEBRUARY 2008
DOI 10.1002/jps